Public Voting and Prosocial Behavior



This readme file includes data and program file explanations.
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++


We have four data files generated from our experiments:
1. Experiment1.dta includes all behavioral data collected in Experiment I.
2. Experiment2.dta includes all behavioral data collected in Experiment II.
3. Figure3.dta includes the data derived from Experiment2.dta that we use to produce Figure3.
4. TableB5.dta includes the data derived from Experiment2.dta that we use to produce Table B5 reported in Appendix B.


++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++


The meaning of each variable contained in Experiment1.dta is explained in the following.

SubjectID: 
The identification of a subject in each session. Notice that because of the setting of our experiment, in the sessions with secret ballots we could not identify who made which decisions, so we put 0s for subjects' IDs in these sessions.

SubjectType: 
1 means A voters, 2 means B voters.

SubjectChoice:
1 means voting for A, 2 means voting for B, 0 means abstention.

Period:
The election period of voting games in the experiment.

WinnerofElection:
It records who the winner is.

SessionNumberinPrivacyType:
It records the session number within a privacy type.

PrivacyType:
It indicates the treatments. We use the same labels in the data file as what we use in the manuscript.

NumberofAvoters:
It indicates the number of A voters in the election.

NumberofBvoters:
It indicates the number of B voters in the election.

ElectionType:
It indicates the type of election.

Sequence:
It indicates the number of Sequence that we explained in the manuscript.


++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++


We use the same labels in Experiment2.dta as what we do in Experiment1.dta except for the following variables.

PrivacyType:
Secret means secret ballots, and Public means public voting.

Treatment:
Short means the first stage of Experiment II is short, Long means the first stage of Experiment II is long.

Session:
It records the session number within a treatment.


++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++


In Figure 3.dta

VotePublic:
It indicates the percentage of voting for "Public Voting."


SubjectType:
It indicates the identity of voters.


Difference:
The difference between the proportion of A wining in secret ballots and the proportion of A winning in public voting 



++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++


In TableB5.dta, we take out the voting data of Periods 11, 17, and 21 in which voters voted for Secret Ballots or Public Voting.



++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++


We also include one Stata program (version: STATA 15.1) for our data analysis and two Matlab (version: 2017b) programs for the estimation of subjects' types based on a mixture model.

The meaning of the key variable in the Matlab codes is the following.

Ob0: The number of votes for abstention.
Ob1: The number of Votes for A.
Ob2: The number of votes for B.

To replicate the results of the mixture model estimation, one should manually enter the results of observations in secretballot.m. One can also copy the following codes and paste them into the matlab file when estimate the values of theta for different voters in different treatments. For any questions, please email Kai Ou at kou@fsu.edu.

The code for Secret Ballots, A voters:


Ob0 = 152;
Ob1 = 196;
Ob2 = 36;


lambda = 10000;

% for A type
%zz1  = exp(u0)+exp(u1)+exp(u2);
prA0 = 0;
prA1 = 0;
prA2 = 1;

% for B type
zz2 = exp(k)+exp(k);
prB0 = exp(k) / zz2;
prB1 = exp(k) / zz2;
prB2 = 0;


The code for Public Voting, A voters:

Ob0 = 52;
Ob1 = 98;
Ob2 = 42;


lambda = 10000;

% for A type
%zz1  = exp(u0)+exp(u1)+exp(u2);
prA0 = 0;
prA1 = 0;
prA2 = 1;

% for B type
zz2 = exp(k)+exp(k);
prB0 = exp(k) / zz2;
prB1 = exp(k) / zz2;
prB2 = 0;
f = - Ob0*log(P0) - Ob1*log(P1) - Ob2*log(P2);
end


 
The code for Secret Ballots, B voters:

Ob0 = 53;
Ob1 = 1;
Ob2 = 74;


lambda = 10000;

% for A type
%zz1  = exp(u0)+exp(u1)+exp(u2);
prA0 = 0;
prA1 = 1;
prA2 = 0;

% for B type
zz2 = exp(k)+exp(k);
prB0 = exp(k) / zz2;
prB1 = 0;
prB2 = exp(k) / zz2;


The code for Public Voting, B voters:

Ob0 = 17;
Ob1 = 7;
Ob2 = 40;


lambda = 10000;

% for A type
%zz1  = exp(u0)+exp(u1)+exp(u2);
prA0 = 0;
prA1 = 1;
prA2 = 0;

% for B type
zz2 = exp(k)+exp(k);
prB0 = exp(k) / zz2;
prB1 = 0;
prB2 = exp(k) / zz2;
